I’m a bit frustrated because I probably won’t be able to do much blogging for the next few weeks, and yet I have a large backlog of issues I’d like to address. I guess the big news today is Paul Krugman’s very generous comments on David Beckworth and I (and by implication the others who have also been pushing the nominal GDP target):
“Market monetarists” like Scott Sumner and David Beckworth are crowing about the new respectability of nominal GDP targeting. And they have a right to be happy.
. . .
At this point, however, we seem to have a broad convergence. As I read them, the market monetarists have largely moved to an expectations view. And now that we’re almost four years into the Lesser Depression, I’m willing, out of a combination of a sense that support is building for a Fed regime shift and sheer desperation, to support the use of expectations-based monetary policy as our best hope.
And one thing the market monetarists may have been right about is the usefulness of focusing on nominal GDP. As far as I can see,the underlying economics is about expected inflation; but stating the goal in terms of nominal GDP may nonetheless be a good idea, largely as a selling point, since it (a) is easier to make the case that we’ve fallen far below where we should be and (b) doesn’t sound so scary and anti-social.
I still believe that the chances of success will be a lot larger if we have expansionary fiscal policy too; but by all means let’s try whatever we can.
That makes me want to take back all the negative Krugman posts I wrote. (Although in fairness, I often called him “brilliant,” and on one occasion argued we’d be much better off if the FOMC had 12 Paul Krugman clones. But I suppose his supporters have noticed the negatives more than the positives.)
I still don’t think I ever denied that the expectations channel was crucial, although I don’t doubt I wrote a couple posts that may have created that impression. But I would like to briefly address his comment about inflation being the theoretically appropriate variable. He may well be right, as I don’t have a magic bullet argument against the mainstream view, but I’ll list a few of my pragmatic arguments:
The mainstream view is that P and Y (prices and output) are the “things” in and of themselves, and P*Y is an ungainly mixture, like a centaur. It treats P and Y equally, even though macro models provide no reason for doing so. So what are my pragmatic arguments?
1. I’ll start with a point I’ve often made, and others like Greg Mankiw have also made. The price index that should be stabilized in the one with sticky prices. To me that means a wage index. Earl Thompson argued for the optimality of wage targeting back in the 1970s, and I continue to see average nominal wages fall when NGDP growth falls sharply. In contrast, the CPI is not a very reliable indicator of excessive monetary tightness, as it’s full of all sorts of flaws. Core CPI does better, but only because it’s much closer to wages. I see NGDP as a sort of proxy for nominal wages.
2. It’s widely thought that low inflation leads to liquidity traps, but the evidence from Japan, China, etc, suggests it’s actually low NGDP growth that leads to liquidity traps. Both had deflation in the late 1990s, but only Japan had low NGDP growth.
3. Inflation appears in many new Keynesian models as a variable used to calculate real interest rates, which then impact AD. But arguably it is the difference between nominal interest rates and nominal GDP growth that is more important. Suppose the SRAS is really flat, so inflation doesn’t rise much with rapid NGDP growth. I believe that if the Fed is able to engineer rapid expected NGDP growth, and nominal rates stay low, that will lead to more investment, even if the real rate doesn’t drop much (because of the flat SRAS.) I think some Keynesians would counter that since you can’t raise inflation substantially without enormous increases in NGDP growth increases (when SRAS is fairly flat), a modest boost in expected NGDP growth is virtually impossible. There’s a sort of “gap” where it’s 1.5% core inflation or 4%, but nothing in between. I think that ignores the fact that monetary policy doesn’t just affect bond yields; it affects the prices of all sorts of assets such as stocks, commodities, commercial real estate, etc, relative to sticky nominal wages. (That’s where my monetarism comes in.) Maybe that’s saying that the “right” price index would include assets. But stock bubbles (i.e. 1987) can give off false signals–so I still prefer NGDP. So replace P and Y with NGDP and hours worked.
4. I also don’t like inflation because I don’t believe the inflation numbers we use correspond to the theoretical concept in NK models. The government says housing costs are up 7.7% in 5 years, Case-Shiller says they are down 32%. Which number best expresses the incentives facing home builders to construct new homes? The Case-Shiller number is actual transaction prices; the BLS number is rental equivalent. Older rental contracts aren’t really prices at all; they are a sort of nominal debt.
5. When using inflation in your model you need to add supply shocks, as higher inflation is only expansionary if it comes from the demand-side. NGDP takes care of that problem—it’s unambiguously demand.
6. There are no solid theoretical foundations for price level theory in a modern economy where hedonics is very important. It’s not just that we’re not good at measuring price changes for computers and consulting services; it’s not even clear what we are trying to measure in theory. Is “a computer” something that yields constant utility? If so, then we need to figure out what utility is. If it’s happiness, and if that’s the theoretical foundation for price level theory, then it means the inflation rate measures the wage increase required to preserve the current level of happiness. In that case, if surveys show people aren’t getting any happier over the decades, then that means RGDP/person is constant. But that’s nonsense. How can inflation be the “theoretically” appropriate variable for these models, if there’s no obvious way to partition computers into prices and output? The only objective fact is the revenue Dell earns from selling computers; the “quantity” is purely arbitrary.
7. Because of point 6, I’m inclined to argue that average hourly wages are the only reasonably objective nominal aggregate that measures a sort of “inflation” (albeit input price inflation.) Yes, total aggregate hours is a bit fuzzy, due to the non-market economy. But the market sector is reasonably well-defined, and gives us something tangible to work with. Set monetary policy to keep average hourly wages growing at a steady rate of 4%, and we’ll be OK. So why don’t I favor a wage target? I’m not sure how comprehensive the wage data actually is, and I also don’t think it’s politically feasible to target wages. It would seem too much like the Fed is trying to hold down wages, which seems unfair. (Even though in theory CEO incomes are also “wages.”)
These arguments are pragmatic, and I could add the one that Krugman alludes to; it’s easier for the Fed to sell a policy that raises the incomes of Americans than one that raises their cost of living. It’s also more honest, as the Fed really is trying to raise NGDP, not pure inflation. That’s been the argument I’ve used more and more often, and indeed I think right now it’s the most powerful argument.
These arguments are all pragmatic, and a few are probably wrong. I doubt that any would cause Krugman to abandon his theoretical framework with its Ps and Ys. But I’m now pretty much a total pragmatist; I no longer believe “inflation” is an actual thing out there in the world, waiting for us to measure it more or less accurately. A famous philosophical pragmatist named Richard Rorty supposed said “truth is what my peers let me get away with saying.” I’d say that right now more and more of my peers are seeing the practical virtues of NGDP targeting. (BTW, almost everyone misunderstood Rorty’s comment.)
PS. Let’s also recall the long and distinguished intellectual tradition of NGDP targeting proposals: Hawtrey, Hayek, McCallum, Taylor, Mankiw, Selgin, and many others.
PPS. Unfortunately, my responses to comments will fall way behind, as I’ll be travelling, then grading.
PPPS. Message to any smart grad students reading this blog: I’m providing the intuition—I’m counting on you to turn it into a rigorous mathematical model that my peers will take seriously.